Parallelizing label propagation for overlapping community detection

Community detection is one of the most important ways that reflect the structure and mechanism beneath the social network. The overlapping communities are more in line with the reality of social network. In the society, the phenomenon of some members shared membership of different communities reflects as overlapping communities in the network. Facing big data network, it is a challenging and computationally complex problem to detect overlapping communities. In this paper, we proposed highly scalable variants of a community detection algorithm with parallelized called Label Propagation with nodes Confidence (PLPAC). We introduced MapReduce to parallelize the algorithm to process the big data and guarantee the efficient of community detection. We implemented the algorithm on real network and artificial network to evaluate the accuracy and speedup of the proposed algorithm. Experiments results on many test datasets illustrated that the improved label propagation method outperforms some existing methods in terms of accuracy and time efficiency.